ECG Signal Analysis for Heart Disease Detection Using Hybrid Machine Learning Techniques

Abstract

Heart disease is one of the leading causes of mortality worldwide, making early detection essential for effective treatment and prevention. Electrocardiogram (ECG) signals provide a non-invasive way to monitor the electrical activity of the heart and identify cardiac abnormalities. This work focuses on developing an intelligent system for accurate detection of heart conditions such as Sleep Apnea, Atrial Fibrillation, and Heart Failure using ECG data. The system utilizes signal processing techniques to remove noise and enhance ECG signal quality, followed by extraction of important features such as RR interval, heart rate, QRS complex, P-wave, T-wave, and PR/QT intervals. In addition to time-domain features, frequency-domain parameters like Power Spectral Density (PSD) and signal energy are incorporated to better capture variations in ECG signals. Machine learning classifiers including Artificial Neural Networks (ANN), K-Nearest Neighbor (KNN), and Support Vector Machine (SVM) are employed to analyze these features and classify cardiac conditions. By combining multiple features and classifiers, the system improves diagnostic accuracy and reduces the chances of misclassification, even in noisy data conditions. This study contributes to the development of a reliable and efficient ECG-based diagnostic system that supports early detection of heart diseases and assists medical professionals in making accurate clinical decisions.

Introduction

The text describes an automated ECG-based heart disease detection system that uses signal processing and machine learning to improve early diagnosis of cardiovascular diseases.

It begins by explaining that ECG is a widely used, non-invasive method for monitoring heart activity, but manual interpretation is slow, error-prone, and difficult due to noise in signals such as baseline drift, motion artifacts, and power-line interference. Traditional methods also rely on limited features, which can reduce diagnostic accuracy.

To address these issues, the proposed system applies advanced preprocessing, feature extraction, and machine learning classification. ECG signals are taken from the PhysioNet database, split into training (70%) and testing (30%) sets.

The processing pipeline includes:

• Preprocessing: Noise removal using filtering and wavelet techniques
• Feature extraction: Time-domain features (R-peaks, RR interval, QRS duration, heart rate, etc.) and frequency-domain features (power spectral density, signal energy)
• Classification: Machine learning models such as ANN, KNN, and SVM, combined into a hybrid classifier for improved accuracy

The system detects conditions like atrial fibrillation, sleep apnea, heart failure, and normal rhythm. Performance is evaluated using accuracy and statistical metrics.

The literature review shows that combining signal processing with machine learning significantly improves ECG-based diagnosis, though challenges remain in feature extraction, noise reduction, and model reliability.

The proposed system follows a structured workflow: ECG data acquisition → preprocessing → feature extraction → normalization → hybrid classification → disease prediction.

Conclusion

Using ECG signal analysis as a reliable and effective method for locating heart disease electrically by the electrical activity of the heart is well-known. The analysis of ECG signals not only gives information concerning the physiological characteristics of an individual’s heart but also enables the identification of problems associated with Sleep Apnea, Atrial Fibrillation, and Congestive Heart Failure. An integration of both Signal Processing Techniques and Machine Learning Techniques allows for more accurate and efficient ECG-based diagnosis. Signal Preprocessing Techniques (e.g., Noise Reduction Techniques) and Wavelet based Denoising Techniques improve the quality of the signal; whereas, Feature Extraction Techniques (time domain characteristics (TD), frequency domain characteristics (FD)) allow for a better analysis because they provide detailed characteristics of ECG Signals. Machine Learning Algorithms (e.g., Artificial Neural Networks (ANN), Support Vector Machine (SVM), K-Nearest Neighbor (KNN)) enable the automated classification of cardiac-related conditions based on characteristics of the cardiac signal. Hybrid Classification Approaches enable improved system performance through using the best features from the combination of the Classification Algorithms and reducing the number of misclassifications. The automated heart disease detection system is a fast, accurate, and reliable method to help healthcare providers make clinical decisions. Automated heart disease detection systems will aid healthcare providers in the early diagnosis and improvement of patient care. As technology advances, ECG-based diagnostic systems will increase in efficiency, availability, and overall use in today’s healthcare system.

References

[1] M. Acharya, A. Oh, Y. Hagiwara, and J. Tan, “ECG-based heart disease detection using hybrid machine learning and signal energy analysis,” IEEE Journal of Biomedical and Health Informatics, 2022. [2] B. Singh and S. Tiwari, “ECG signal denoising using wavelet transform and performance evaluation,” in Proc. IEEE Int. Conf. Signal Processing and Integrated Networks (SPIN), 2018, pp. 1–6. [3] A. L. Goldberger et al., “PhysioBank, PhysioToolkit, and PhysioNet: Components of a new research resource for complex physiologic signals,” Circulation, vol. 101, no. 23, pp. e215–e220, 2000. [4] S. Osowski, L. T. Hoai, and T. Markiewicz, “Support vector machine-based expert system for reliable heartbeat recognition,” IEEE Transactions on Biomedical Engineering, vol. 51, no. 4, pp. 582–589, 2004. [5] P. Rajpurkar et al., “Cardiologist-level arrhythmia detection with convolutional neural networks,” IEEE Journal of Biomedical and Health Informatics, vol. 21, no. 5, pp. 1–10, 2017. [6] U. R. Acharya et al., “Automated diagnosis of congestive heart failure using deep convolutional neural network with ECG signals,” Applied Intelligence, vol. 49, pp. 16–27, 2019. [7] R. Elantary, “Artificial intelligence in electrocardiography,” IEEE Reviews in Biomedical Engineering, 2025. [8] S. Satija, B. Ramkumar, and M. Manikandan, “A review of signal processing techniques for electrocardiogram signal quality assessment,” IEEE Reviews in Biomedical Engineering, vol. 11, pp. 36–52, 2018.

Copyright

Copyright © 2026 Dr. G. Krishna Reddy, Katakam Meghana, Kota Jahnavi, Aremarati Lakshmi Kavya, Endravath Swapna . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.